PDMP sensitizes neuroblastoma to paclitaxel by inducing aberrant cell cycle progression leading to hyperploidy Free

Paclitaxel is an antitumor drug that binds to β-tubulin in microtubules and promotes αβ-tubulin assembly. By permanent activation of tubulin, paclitaxel stabilizes microtubules. Paclitaxel and other microtubule poisons have in common that they suppress dynamic behavior of microtubules and therefore have their strongest effects on the mitotic spindle microtubules during mitosis (1). Mitotic spindle microtubules are more dynamic than interphase microtubules, and their dynamic behavior includes “treadmilling” and “dynamic instability,” required for correct interaction with alignment and segregation of chromatids. Paclitaxel can thus interfere with correct mitotic progression and induce cell cycle blockage (2). The effects of paclitaxel are concentration dependent (3). At low concentrations, bipolar spindles still form and mitosis can progress until metaphase/anaphase transition. Then, mitotic blockage occurs apparently due to suppressed microtubule dynamics, which may result in a permanent activation of the mitotic checkpoint and subsequent induction of apoptosis (4). Alternatively, the block is transient, and progression to a G₁-like stage occurs (mitotic slippage). As this usually involves an aberrant exit from mitosis with abnormal chromatid segregation, multinucleated cells are formed which are prone to engage in apoptosis (5). At high paclitaxel concentrations, bipolar spindle formation is affected possibly due to an additional effect of paclitaxel on centrosome function (3). This results in monopolar and multipolar spindles, permanent activation of the mitotic checkpoint, and subsequent induction of apoptosis.

d,l-Threo-1-phenyl-2-decanoylamino-3-morpholino-1-propanol (t-PDMP) is a well-known inhibitor of sphingolipid biosynthesis (6). The formation of glucosylceramide is drastically inhibited by t-PDMP, and a concomitant accumulation of its endogenous sphingolipid precursor ceramide may occur. Ceramide is a key player in the regulation of apoptosis, making it an attractive target for cancer therapy (7). Indeed, inhibitors of glucosylceramide synthase (GCS) have been used to reverse drug resistance and it is generally assumed that this is based on their ability to induce and/or increase ceramide accumulation in combination with cytostatics, leading to apoptosis (8). Recently, however, the involvement of GCS in drug sensitization by PDMP has been questioned (ref. 9; for reviews on this issue, see refs. 10, 11). It was observed that iminosugars, which are potent inhibitors of GCS, do not chemosensitize multidrug-resistant tumor cells to cytostatics (9, 12). The notion thus arises that chemosensitization achieved by PDMP may not involve GCS but instead is caused by an alternative mechanism.

In this study, we sensitized Neuro-2a murine neuroblastoma cells to the microtubule-stabilizing agent paclitaxel, using various GCS inhibitors, including various PDMP analogues and N-butyldeoxynojirimycin (NB-dNJ; ref. 13). The Neuro-2a model was chosen because we have shown previously that these cells are susceptible to PDMP-mediated sensitization to paclitaxel (14). Here, we show that synergistic inhibition by PDMP and paclitaxel of the viable cell number increase is not based on a high rate of apoptosis but involves a novel action of PDMP (i.e., induction of hyperploidy). Coincubation of paclitaxel with low concentrations of PDMP resulted in aberrant mitosis, which in combination with mitotic slippage lead to hyperploidy after multiple cell cycles. This effect was specific for PDMP but was not related to GCS inhibition and/or ceramide accumulation. Instead, PDMP acted by itself together with paclitaxel to inhibit cyclin-dependent kinases (CDK) and thus affected cell cycle.

In conclusion, this article provides evidence for a novel, GCS- and ceramide-independent action of PDMP underlying its chemosensitizing effects at low concentrations (i.e., synergistic induction of hyperploidy rather than apoptosis). This may provide the basis for resolving the discrepancy regarding differential effects of various GCS inhibitors as chemosensitizers.

The murine anti-giantin antibody was a kind gift from Dr. Hauri (Department of Pharmacology, University of Basel, Basel, Switzerland). The murine neuroblastoma cell line Neuro-2a was purchased from the American Type Culture Collection (Manassas, VA). Cell Death Detection ELISA^PLUS kit and In situ Cell Death Detection kit (fluorescein) were from Roche Diagnostics Corp. (Indianapolis, IN). l-[U-¹⁴C]serine (150 mCi/mmol) and [γ-³²P]ATP were purchased from Amersham Pharmacia Biotech UK Ltd. (Buckinghamshire, United Kingdom). All PDMP analogues were purchased from Matreya LLC (Pleasant Gap, PA). NB-dNJ was from Biomol (Plymouth Meeting, PA). ISP-1, roscovitin, rabbit anti-γ-tubulin antibody, Hoechst 33528, 4′,6-diamidino-2-phenylindole, and 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) were from Sigma-Aldrich (St. Louis, MO). Anti-CDK2 antibody (D-12) and Protein A/G Plus-agarose beads were from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). Agarose-bound p13^suc1 was obtained from Upstate (Lake Placid, NY). Paclitaxel was purchased from Calbiochem, Merck Biosciences Ltd. (Nottingham, United Kingdom).

The murine neuroblastoma cell line Neuro-2a was grown as described (14).

The MTT assay was done as described (12) using 10³ cells per well.

After various treatments, Neuro-2a cells were trypsinized. The cells were washed twice with HBSS and their protein content was determined according to Smith et al. (15). Protein (2.5 μg) was incubated in lysis buffer as provided in the Cell Death Detection ELISA^PLUS kit. After centrifugation (10 minutes, 200 × g), the relative enrichment of DNA-histone complexes in the cytoplasmic fraction was measured with an ELISA according to the manufacturer's manual.

Cells were grown on glass coverslips and exposed to several treatments. On fixation with 4% paraformaldehyde, the In situ Cell Death Detection kit was used to stain apoptotic (terminal deoxynucleotidyl transferase–mediated dUTP nick end labeling–positive) cells. Analysis of the samples was done on an Olympus Provis AX70 epifluorescence microscope (Olympus Optical Co. GmbH, Hamburg, Germany).

Flow cytometric analysis was used for cell cycle analysis of propidium iodide–stained cells. Cells were plated in 25-cm² plates and allowed to adhere for 4 hours before treatment with various concentrations of paclitaxel, t-PDMP, and/or ISP-1. Cells were harvested by trypsinization, washed with HBSS, and fixed with 70% ethanol on ice for 1 hour. Fixed cells were washed thrice with PBS, resuspended in PBS containing RNase A (500 μg/mL), and incubated at 37°C for 30 minutes. Cells were incubated with propidium iodide (50 μg/mL) at least 30 minutes before analysis by flow cytometry. Flow cytrometric analysis was carried out using an Elite flow cytometer (Beckman Coulter, Miami, FL) and ModFit LT 3.0 software (Verity Software House, Inc., Topsham, ME).

Immunofluorescence microscopy was carried out essentially as described (16). Primary antibodies (incubation at 4°C, overnight) were rabbit anti-γ-tubulin (1:500) and mouse anti-giantin (1:500). Secondary antibodies (incubation at room temperature, 2 hours) were rhodamine (TRITC) goat anti-mouse antibody and fluorescein (FITC) goat anti-rabbit antibody. Hoechst 33528 was applied during the secondary antibody incubation. Samples were analyzed by confocal laser scanning microscopy (TCS Leica SP2 Confocal Laser Scanner Microscope, Leica, Heidelberg, Germany).

To determine the mitotic index, Neuro-2a cells were grown on coverslips, incubated in the presence of paclitaxel and/or t-PDMP, fixed with methanol (−20°C, 5 minutes), stained with 4′,6-diamidino-2-phenylindole, and analyzed by fluorescence microscopy. At least 300 cells were counted per condition in triplicate. Cells with condensed chromosomes were judged as mitotic.

Sphingolipid pools were analyzed as described (12).

CDK1 activity was determined according to Lee et al. (17), with minor protocol modifications: the AEBSF was replaced by 1 mmol/L phenylmethylsulfonyl fluoride in the TBSN buffer, cells were passed 10 times through a 25-gauge needle before ultracentrifugation, and TBMB buffer supplemented with 2 μg histone H1, 20 μmol/L ATP, and 5 μCi [γ-³²P]ATP was used as the kinase cocktail. To determine CDK2 activity, cells were washed with HBSS and scraped in 1 mL NP40 buffer [150 mmol/L NaCl, 1% NP40, 50 mmol/L Tris, 1 mmol/L phenylmethylsulfonyl fluoride, protease inhibitors (pH 8.0)]. Cell lysates were passed 10 times through a 25-gauge needle and centrifuged (10 minutes, 12,000 × g). Affinity purification was done on supernatants containing 500 μg protein using anti-CDK2 antibody combined with Protein A/G Plus-agarose beads. After a pre-clear, antibody incubation, and antibody incubation in the presence of beads (all 1 hour, 4°C), beads were washed twice with radioimmunoprecipitation assay buffer [NP40 buffer supplemented with 0.5% sodium deoxycholate, 0.1% SDS (pH 8.0)] and twice with kinase buffer [20 mmol/L Tris-HCl, 10 mmol/L MgCl₂, 1 mmol/L DTT, protease inhibitors (pH 7.4)]. CDK2 reactions were done in 50 μL kinase buffer supplemented with 2 μg histone H1, 20 μmol/L ATP, and 5 μCi [γ-³²P]ATP (30 minutes, 37°C) and stopped by adding 5× Laemmli sample buffer. Samples were resolved on a 10% SDS-PAGE, which was fixed and exposed to a Phosphor Screen (Molecular Dynamics/Amersham Biosciences, Buckinghamshire, United Kingdom). Bands of phosphorylated histone were quantified using a Storm PhosphorImager (Molecular Dynamics/Amersham Biosciences) and Scion Image Beta 4.0.2 software (Scion Corp., Frederick, MD).

Results are mean ± SD of at least three independent experiments. Statistical analysis was done with the Student's t test, considering P < 0.05 significant.

Cotreatment of Neuro-2a cells with up to 10 μmol/L t-PDMP and up to 50 nmol/L paclitaxel resulted in a synergistic inhibition of viable cell number increase compared with treatment with t-PDMP or paclitaxel alone (Fig. 1A). After 4-hour treatment with the combination of 50 nmol/L paclitaxel and 3 μmol/L t-PDMP, cells displayed an organelle morphology that was indicative of an early mitotic block. This included condensed DNA in a donut-like shape with centrosomes positioned in the center of the condensed DNA, whereas the Golgi was fragmented (Fig. 1C). In control cells, the centrosomes displayed a perinuclear localization, whereas the Golgi was intact and in close proximity to the centrosomes (Fig. 1E). In accordance, the mitotic index was synergistically increased in cells treated with both paclitaxel and t-PDMP (Fig. 1B) and there was a pronounced effect on the cell cycle distribution with a large G₂-M peak after 6 hours of treatment (Table 1A). These cells did not resume normal cell division. Instead, paclitaxel/t-PDMP-treated cells became hyperploid within 24 hours. This was evident from the flow cytometric analysis of cell cycle distribution (Fig. 2A and B), showing that 50% of the cells had become hyperploid following 24 hours of treatment (Table 1B). After 48 hours of treatment, even 8N and 16N DNA peaks could be distinguished in addition to 2N and 4N peaks (Fig. 2C). Morphologically, hyperploid cells were characterized by asymmetrically dividing nuclei, which surrounded an often abnormally large Golgi and centrally located centrosomes (Fig. 1D compared with control, Fig. 1F).

Separately, paclitaxel and t-PDMP did not induce formation of hyperploid cells (Table 2). Clearly, both agents are required to induce this effect. t-PDMP alone had no effect on viable cell number (Fig. 1A, Y axis), mitotic index (Fig. 1B), organelle morphology (Fig. 1G), and cell cycle distribution (Table 1A and B). Paclitaxel alone did cause an early mitotic block (Fig. 1G) paralleled by a significant early increase of the mitotic index (Fig. 1B) as well as an increase in the number of G₂-M-phase cells after 6 hours (Table 1A). However, the fraction of G₂-M-phase cells had normalized to control levels after 24 hours of paclitaxel treatment (Table 1B). At this time point, cells with morphologic characteristics of an early mitotic block were no longer observed (Fig. 1G). Thus, paclitaxel induced a transient mitotic arrest in Neuro-2a cells, but the cells had recovered completely after 24 hours of treatment.

To exclude that the reversal of mitotic arrest was due to inactivation of paclitaxel during the 24-hour incubation, the following experiments were done. After 24 hours of paclitaxel treatment, the medium was replaced with either fresh medium or fresh paclitaxel-containing medium and cells were incubated for an additional 48 hours. This situation was compared with 72-hour continuous incubation with paclitaxel (Table 2). These experiments showed that the paclitaxel-induced arrest was indeed transient, because there was no further inhibition of viable cell number increase on addition of fresh paclitaxel (cf. 24-hours paclitaxel to 48-hour paclitaxel with 72-hour paclitaxel). It should be noted that a slight inhibition of viable cell number increase resulted from the incubation medium shift by itself as indicated in Table 2 (bottom).

Paclitaxel and t-PDMP might act independently with additive effect. To investigate this, the effect of paclitaxel/t-PDMP versus t-PDMP alone on viable cell number was studied in paclitaxel preincubated (24-hour) cells. Paclitaxel preincubated (24-hour) cells had overcome the temporary mitotic block and were resistant to a second challenge (48 hours) with paclitaxel (Table 2). Subsequent treatment with both paclitaxel and t-PDMP, however, further inhibited viable cell number increase from 53% to 23% (Table 2). Replacing paclitaxel-containing medium with t-PDMP-containing medium after 24 hours resulted in a viable cell number (57%) similar to the condition in which plain medium was added (53%; Table 2). In the case of the reversed order of incubation [i.e., a preincubation with t-PDMP (24 hours) followed by a 48-hour incubation with paclitaxel], the viable cell number was 66% (n = 2). Taken together, these data indicate that the simultaneous presence of paclitaxel and t-PDMP is required for the formation of hyperploid cells.

The cell cycle distribution analysis indicated a low level of apoptosis in paclitaxel/t-PDMP-treated cells. To quantify this, apoptosis was determined with both a cytoplasmic histone-associated DNA ELISA and a terminal deoxynucleotidyl transferase–mediated dUTP nick end labeling assay (Table 3). Control cells displayed a very low level of apoptosis. Although in both assays the level of apoptosis increased in the case of paclitaxel/t-PDMP treatment, the apoptosis end-point value remained low (Table 3). Therefore, these experiments support the conclusion that formation of hyperploid cells and not apoptosis is at the basis of the paclitaxel/t-PDMP-induced synergistic inhibition of viable cell number increase of Neuro-2a cells.

t-PDMP is known for its inhibitory action on GCS. This may lead to increased levels of ceramide, which may be further enhanced by paclitaxel, if the latter induces de novo sphingolipid biosynthesis. Therefore, we studied whether (a) GCS and/or (b) ceramide accumulation were involved in paclitaxel/t-PDMP-induced hyperploidy.

(a) We tested three other GCS inhibitors [d,l-erythro-1-phenyl-2-decanoylamino-3-morpholino-1-propanol (e-PDMP), d,l-threo-1-phenyl-2-palmitoylamino-3-pyrrolidino-1-propanol (t-PPPP), and NB-dNJ]. All three inhibitors had different effects on the viable cell number in the presence of paclitaxel. e-PDMP (Fig. 3A) strongly resembled t-PDMP (Fig. 1A) in synergistic action. t-PPPP did not act synergistic with paclitaxel but inhibited viable cell number increase independent of paclitaxel at 1 and 3 μmol/L concentration (Fig. 3B). NB-dNJ did not affect the viable cell number at all (Fig. 3C). Like t-PDMP, e-PDMP induced hyperploidy in combination with paclitaxel, but t-PPPP and NB-dNJ did not, as determined by fluorescence-activated cell sorting analysis and microscopy (Table 4). The latter two inhibitors were the most effective GCS inhibitors both alone and in combination with paclitaxel (Table 5). In fact, e-PDMP seemed to have hardly any inhibiting effect on GCS, whereas the intermediary effect of t-PDMP was not significant. Thus, the two most effective GCS inhibitors, t-PPPP and NB-dNJ, did not synergize with paclitaxel to induce hyperploidy and inhibit viable cell number increase. The least effective GCS inhibitor, e-PDMP, did synergize with paclitaxel. This rules out GCS inhibition as a cause for hyperploidy.

(b) All four GCS inhibitors, in combination with paclitaxel, resulted in a significant increase in ceramide level (Table 5), whereas only two of the four inhibitors synergize with paclitaxel. Thus, it seems that induction of hyperploidy and synergistic inhibition of viable cell number increase is not correlated with increased formation of ceramide. Under these conditions, the possibility still exists that a high accumulation of ceramide, as in the case of t-PDMP or e-PDMP, combined with paclitaxel, is required. In the case of t-PPPP and NB-dNJ, enhanced sphingomyelin levels were observed, suggesting that excess ceramide was efficiently converted to sphingomyelin (Table 5). However, further experiments ruled out ceramide accumulation as the cause of hyperploidy, as became apparent when ISP-1 was used to inhibit ceramide biosynthesis. ISP-1 (0.5 μmol/L) was a very effective inhibitor in Neuro-2a cells, causing 88 ± 5% (n = 3) reduction of sphingolipid biosynthesis. In spite of this reduction, ISP-1 did not prevent hyperploidy (Table 4) and Neuro-2a cells were not protected by ISP-1 from the synergistic inhibition of viable cell number increase (Fig. 3D). The latter data show that ISP-1 is slightly toxic by itself. Indeed, an increase in apoptosis due to ISP-1 treatment was observed (Table 3). Paclitaxel, t-PDMP, or the combination did not further increase the level of apoptosis (Table 3). It is important to note that ISP-1 by itself or in combination with either paclitaxel or t-PDMP did not induce hyperploidy (Table 4).

Our data presented thus far indicate a ceramide/GCS-independent effect of paclitaxel/t-PDMP on cell cycle progression, resulting in the formation of hyperploid cells. We next studied whether modulation of the activity of CDKs could be the underlying mechanism, because we have observed that roscovitin, an established inhibitor of CDKs, also synergized with paclitaxel to inhibit viable Neuro-2a cell number increase (data not shown). Direct evidence for their involvement came from the measurement of the relative kinase activities of CDK1 and CDK2. CDK activities were measured (a) after pretreatment for 24 hours with paclitaxel and/or t-PDMP and (b) with paclitaxel and t-PDMP present only during the kinase reaction. This allowed us to discriminate between direct effects of paclitaxel and/or t-PDMP on kinase activity and indirect (long-term) effects. Both CDK1 and CDK2 activities were indirectly and synergistically inhibited by the paclitaxel/t-PDMP combination, whereas neither paclitaxel alone nor t-PDMP alone were inhibitory (Fig. 4A and C). There were no direct effects of t-PDMP and/or paclitaxel on either CDK1 or CDK2 activity in contrast to the inhibitory effect of roscovitin (Fig. 4B and D). Thus, synergistic inhibition of both CDK1 and CDK2 activities by paclitaxel/t-PDMP matched synergistic induction of hyperploidy and inhibition of viable cell number increase, whereas paclitaxel alone or t-PDMP alone did neither inhibit CDK activities nor induce hyperploidy.

PDMP and/or its analogues like PPPP have been used in numerous studies to sensitize (multidrug-resistant) tumor cells to cytostatics (10). It is generally assumed that this sensitizing effect is caused by the inhibitory effect of PDMP on GCS, thereby resulting in increased ceramide levels. Ceramide in turn would reduce cell survival by induction of apoptosis. According to this view, PDMP strengthens the effect of the cytostatic, which by itself induces ceramide accumulation through activation of de novo ceramide biosynthesis or alternatively through activation of a sphingomyelinase.

Recently, iminosugars, such as N-nonyl-deoxygalactonojirimycin and N-butyl-deoxy(galacto)nojirimycin, have been used as alternative GCS inhibitors in multidrug-resistant tumor cell sensitization studies. These compounds, which seem to be more specific and less toxic, did not sensitize multidrug-resistant tumor cells to cytostatics (9, 12). Norris-Cervetto et al. (9) concluded from their studies that the effect of PDMP on drug resistance could not be explained by inhibition of GCS alone. However, there are no studies providing a mechanism unrelated to GCS inhibition by which PDMP could chemosensitize multidrug-resistant tumor cells.

In the present study, we show that PDMP-mediated chemosensitization of Neuro-2a cells is dissociated from both GCS inhibition and ceramide accumulation. (a) The most effective GCS inhibitors of this study, t-PPPP and NB-dNJ, did not synergize with paclitaxel to inhibit viable cell number increase, with NB-dNJ having no effect at all on cell viability. Of the GCS inhibitors tested, the synergistic effect was specific for PDMP, occurring with both t-PDMP and e-PDMP. In this respect, a similar dichotomy was observed between t-PDMP and t-PPPP in HepG2 cells: unlike t-PDMP, t-PPPP did not potentiate apoptosis in doxorubicin-treated HepG2 cells (18). (b) The synergistic combination of t-PDMP or e-PDMP and paclitaxel resulted in the largest accumulation of ceramide. These results agree with results obtained previously in our laboratory, which showed a correlation between t-PDMP-mediated sensitization of Neuro-2a cells to the microtubule-perturbing agents paclitaxel and vincristine and long-term ceramide accumulation (14). However, in the present study, we show that the nonsynergistic combinations of t-PPPP or NB-dNJ and paclitaxel also resulted in a significant ceramide increase. More importantly, we show that a complete abrogation of sphingolipid synthesis in ISP-1-treated cells did not affect synergism, showing its independence of ceramide accumulation.

We cannot exclude the possibility that ceramide accumulation contributed to the minor degree of apoptosis observed, which would be consistent with a previous report pointing to apoptosis as the basis for PPPP-mediated chemosensitization in leukemia cells (19). However, apoptosis is not the main feature contributing to inhibition of viable cell number increase of paclitaxel/t-PDMP-treated Neuro-2a cells. Instead, a novel action of PDMP (i.e., induction of hyperploidy through aberrant cell cycle progression in paclitaxel-arrested cells) is at the basis of the synergism between paclitaxel and PDMP, which are both required at the same time to produce the effect. Paclitaxel alone induced a mitotic delay, from which the cells were able to escape and complete cytokinesis apparently normal within 24 hours. Paclitaxel/t-PDMP-treated cells also displayed a mitotic block; however, these cells escaped in a process known as mitotic slippage (i.e., by entering G₁ without complete chromosome segregation and cell division). On mitotic slippage, cells with a proper G₁ checkpoint are arrested; subsequently, apoptosis is triggered in a p53- or Rb-dependent way, preventing cells from reentering S phase (20, 21). Fluorescence-activated cell sorting analysis showed that paclitaxel/t-PDMP-treated Neuro-2a cells underwent multiple rounds of DNA synthesis after mitotic slippage, resulting in the formation of hyperploid cells. This is indicative of an impaired G₁ checkpoint, which could be an inherent property of Neuro-2a cells. The low levels of apoptosis after 24 hours of treatment support the conclusion that these cells are able to enter a next round of DNA synthesis on mitotic slippage. It is likely that the cell cycle effects of paclitaxel/t-PDMP only became apparent due to the inherent impaired G₁ checkpoint of Neuro-2a cells. In other cell types, this effect may have gone unnoticed due to subsequent massive apoptosis. Thus, the Neuro-2a model turns out to be a good model to track early, upstream effects of paclitaxel/t-PDMP that ultimately lead to inhibition of viable cell number increase, be it via an aberrant cell cycle or subsequent apoptosis. Synergy between paclitaxel and t-PDMP and aberrant cell cycle progression leading to hyperploidy did not specifically occur in Neuro-2a cells but has broader significance, because these two effects were also observed in Chinese hamster ovary cells as indicated by MTT analysis and microscopy, respectively (data not shown).

We provide evidence for a GCS-independent molecular mechanism of chemosensitization by PDMP (i.e., through indirect modulation of CDKs). t-PDMP-induced cell cycle arrest at G₁-S and G₂-M transitions has been observed in NIH-3T3 cells by Rani et al. (22). In their study, relatively high concentrations of t-PDMP (50 μmol/L) were used, which directly inhibited both CDK1 (p34^cdc2) and CDK2 activity. In our study, much lower concentrations of t-PDMP were used, which by themselves did not inhibit viable cell number increase of Neuro-2a cells. We show that t-PDMP at this concentration did not directly inhibit CDK1 or CDK2 activity. However, in combination with paclitaxel, a synergistic inhibition of both CDK1 and CDK2 activities was observed. These effects are most likely involved in induction of hyperploidy and inhibition of viable cell number increase given the synergistic nature of both responses.

In summary, we have shown that the paclitaxel/t-PDMP combination induces hyperploidy in Neuro-2a murine neuroblastoma cells, involving abnormal chromatid segregation and aberrant exit from mitosis (mitotic slippage) to a G₁-like stage of 4N cells. Due to an apparent dysfunction of the G₁ checkpoint, cells subsequently progress through multiple aberrant cell cycles to become hyperploid. This leads to a synergistic inhibition of viable cell number increase of Neuro-2a cells. In accordance with aberrant cell cycle progression and dysfunctional cell cycle checkpoints, both CDK1 and CDK2 turn out to be indirect targets of the paclitaxel/t-PDMP combination. Our study provides a novel basis for the understanding of PDMP-mediated sensitization of tumor cells to cytostatics, which is independent of its inhibitory effect on GCS and therefore cannot be reproduced by other (and more specific) GCS inhibitors. These effects of PDMP may well be restricted to cytostatics, which affect microtubule dynamics, because we have shown previously that synergy is observed with two different microtubule-affecting agents (paclitaxel and vincristine) but not with the topoisomerase II inhibitor VP16/etoposide (14). We therefore hypothesize that the effects are related to microtubule dynamics involved in cell cycle progression, such as proper chromatid separation. Even if unrelated to sphingolipid biology, the cell biological effects of PDMP (especially at low concentrations) and the understanding of the underlying molecular mechanism may pave the way for future antitumor treatments based on chemosensitization to (microtubule-affecting) cytostatics.